RoofLang: Enabling AI-Driven Architecting of LLM Inference Systems
This work argues that a general workload representation, a verifiable mutation space, and an implementation-independent evaluator are required to enable the AI-driven LLM inference system architecting loop, and presents the RoofLang domain-specific language (DSL) that provides these features.
Ziyue Yang, Yu-Ting Jiang, L. Qu et al.
· 0 citations